The team, led by Wenchang Feng and Jixiong Zhang with colleagues including Meng Li, Zhibo Cui, Zhangyu Li, and Yuzhe Zhang, set out to tackle two intertwined environmental burdens. Decades of high-intensity coal extraction have left vast stockpiles of coal gangue and fly ash that occupy land, risk spontaneous combustion and landslides, and can leach contaminants into soil and groundwater. Meanwhile, groundwater dissolving soluble minerals in coal-bearing strata produces mine water so laden with salts that discharging it untreated would salinize soils and pollute waterways. Desalination plants at the mines can strip the salts out, but the process concentrates them into a brine whose disposal is both difficult and expensive, with evaporative crystallization currently the dominant — and energy-hungry — fallback.

The researchers’ proposal was elegantly simple: let the brine do useful work. They prepared a high-salinity concentrated mine water-based coal-based solid waste cemented backfill, abbreviated HW-CB, using crushed coal gangue as coarse aggregate, fly ash as fine aggregate and supplementary cementitious material, and Portland cement as the binder. A parallel series made with tap water, TW-CB, served as the control. The brine itself came from the treatment plant at Jinqiao Coal Mine in Jining City, and its chemistry was striking: sulfate concentrations 313 times those of tap water, sodium 244 times higher, potassium nearly 40 times higher, and chloride about 17 times higher, all at a near-neutral pH of 7.52.

Flow behavior proved to be the first sensitive variable. As the solid content of the slurry rose from 74 to 82 percent by weight, slump declined gently at first and then collapsed — the material held a workable 278 to 282 millimeters up to 76 percent solids, then lost 28, 60, and 76 millimeters for each successive 2 percent increase. The team attributes the cliff-like drop to a threshold near 78 percent solids, beyond which free water can no longer coat every particle, friction and particle bridging take over, and flowability deteriorates drastically. Replacing fly ash with more cement improved slump in both water types, because the fly ash’s free calcium oxide slakes rapidly and consumes water within minutes, while cement’s calcium is locked in slow-hydrating silicate minerals. Yet at every cement dosage, the saline mix flowed slightly worse than its tap-water twin, since the dense ion soup suppresses electrostatic repulsion between particles and accelerates the formation of early hydration products that thicken the slurry.

Strength told a more surprising story. The 28-day unconfined compressive strength of the saline backfill rose linearly with solid content, gaining roughly 0.86 megapascals for every 1 percent increase in solids, with a coefficient of determination of 0.997 — climbing from 2.9 to 9.6 megapascals across the tested range. More intriguing was the interplay between cement dosage and salinity. At cement proportions of 6 percent or higher, the saline mix outperformed the tap-water mix at every age, with 3-day, 7-day, and 28-day strengths all higher. Below about 4 percent cement, the pattern inverted: early strengths were still boosted, but the 28-day strength fell below the tap-water control, with the 4-to-6 percent range acting as a transition interval.

Microscopy and diffraction explained why. X-ray diffraction revealed that the intensity of the ettringite peak tracked the strength results exactly. When calcium was plentiful, the sulfate, chloride, sodium, and potassium ions in the brine continuously drove ettringite formation while accelerating the dissolution of the aluminate phase, producing a dense microstructure in which well-developed, thick ettringite crystals interlocked into a three-dimensional skeleton and abundant calcium silicate hydrate gel filled the pores. Scanning electron microscopy showed the saline mix riddled with only small, disconnected pores, whereas the tap-water counterpart contained large connected voids and finer, shorter crystals. Thermogravimetric analysis sealed the case: mass loss between 100 and 200 degrees Celsius, which reflects bound water in the hydration products, ranked in precisely the same order as the measured strengths.

The mechanism behind the low-cement failure is a classic case of too much of a good thing. With insufficient calcium available, the brine’s acceleration caused hydration products to form so quickly that they encapsulated the unhydrated particles, sealing off the pathways later reactions needed. The result was short, sparse ettringite, unfilled voids, and a weaker final structure than the slower, steadier tap-water hydration achieved. In effect, the high-salinity environment acts as a throttle on the hydration clock — helpful when there is enough fuel to sustain the reaction, harmful when the fuel runs out early.

The economics may prove as compelling as the chemistry. At Jinqiao Coal Mine, the current zero-liquid-discharge scheme combining membrane concentration with evaporation and crystallization costs about 15.83 yuan per cubic meter of treated water, with throughput that can exceed 10,000 cubic meters per day during high-inflow periods. If the concentrated brine from the reverse-osmosis stage is diverted into backfill production instead, the downstream evaporation and mixed-salt disposal steps can be skipped, cutting the unit treatment cost by roughly 8 to 9 yuan per cubic meter — a reduction of more than 50 percent. On an annual treatment volume of 120,000 cubic meters, that translates into savings of roughly 960,000 to 1,080,000 yuan, before counting the simplified flowsheet and reduced capital costs.

The researchers are careful to note the boundaries of their findings. The backfill is designed for unreinforced goaf filling, so the brine’s chloride content poses no steel-corrosion concern in this application, but long-term strength evolution at 90 or 180 days, durability under sustained saline exposure, the effects of varying brine compositions, and full-scale field trials all remain open questions. Still, the core result stands: with the cement dosage tuned above the critical threshold, a waste stream that mines once paid to destroy becomes an active ingredient in the very material that stabilizes the voids coal mining leaves behind — turning two liabilities into one structural asset.

Subject of Research: Using high-salinity concentrated mine water as mixing water in coal gangue and fly ash cemented backfill and its effects on flow, strength, and hydration

Article Title: Performance and hydration mechanism of backfill materials prepared by coal-based solid waste and high-salinity concentrated mine water

Article References: Feng, W., Zhang, J., Li, M., Cui, Z., Li, Z., & Zhang, Y. (2026). Performance and hydration mechanism of backfill materials prepared by coal-based solid waste and high-salinity concentrated mine water. Case Studies in Construction Materials, 25, Article e06558. https://doi.org/10.1016/j.cscm.2026.e06558

Image Credits: AI Generated

DOI: 10.1016/j.cscm.2026.e06558

Keywords: cemented backfill, coal gangue, fly ash, high-salinity mine water, ettringite, hydration mechanism, unconfined compressive strength, solid waste utilization, mine water treatment, zero-liquid discharge, Portland cement, X-ray diffraction

The research team, led by Yongjing Ye and Keqing Li, carried out their investigation at Longshou Mine, where the drift cross-section is hexagonal, with a sidewall width of 6 meters, a floor width of 4 meters, and a height of 5 meters. The load-bearing layer of backfill above the drift is 3 meters thick. Before filling, crews place triangular steel trusses on the drift floor and lay steel mesh above them, lapping the mesh continuously along the floor and up both sidewalls. The longitudinal bars of the mesh, spaced at 300 millimeters, and the main bars of the trusses act together as the tensile reinforcement of the resulting composite slab.

To model this structure, the researchers turned to a classic tool of reinforced concrete theory: the transformed section method. Because steel is far stiffer than cemented backfill, with an elastic modulus ratio of roughly 122 in this case, each bar’s cross-sectional area can be converted into an equivalent area of backfill. This allows the neutral axis position, the moment of inertia, and the stress distribution of the whole section to be calculated with standard beam mechanics. Before cracking, the model treats the backfill as a homogeneous, linear-elastic material obeying the plane-section assumption; after cracking, it assumes the fractured tension zone carries no tensile force at all, leaving the steel to act alone. That second assumption is deliberately conservative, meaning the real capacity of a cracked roof is likely higher than the model predicts.

The roof itself is idealized as a simply supported beam spanning the drift width, loaded by a uniformly distributed overburden pressure. The authors are candid about the limits of this idealization: real drift ends receive some rotational restraint from surrounding rock and adjacent fill, so the simply supported model may slightly underestimate stiffness and overestimate deflection. Self-weight is likewise excluded from the capacity calculations because it is identical for every reinforcement scheme considered, making the results valid comparative indices rather than absolute design loads.

Validation came from an ambitious field monitoring program. Because embedding and fixing sensors in soft backfill is notoriously difficult, the team designed a dedicated fixing device and installed pressure cells at elevations of 1, 2, and 3 meters above the drift floor, concrete strain gauges at four elevations, and rebar stress gauges welded directly into the steel mesh and trusses. All instruments sat at midspan, and convergence monitoring sections were arranged along several drifts to track roof subsidence as mining advanced. The results showed the overburden load on the roof’s load-bearing layer holding steady between 0.11 and 0.16 megapascals, with strain flipping from compression near the top of the section to progressively larger tension near the bottom, exactly the pattern a bending beam should produce.

The comparison between theory and measurement was instructive. The model predicted an initial cracking capacity of 0.30 megapascals, comfortably above the measured 0.14 megapascals, confirming the roof stayed uncracked throughout monitoring, which matched field observation. Predicted strains at 0.5, 1.0, and 1.5 meters above the floor matched measured trends with errors of only 18, 38, and 2 microstrain. At 2.0 meters, however, the measured compressive strain of 632 microstrain vastly exceeded the predicted 58, a discrepancy the authors attribute to local stress concentrations, pressure arching, material heterogeneity, and the fact that a single gauge records a local response while the model describes an average. Rebar gauge readings told a similar story: theoretical forces of 6.3 and 6.7 kilonewtons fell within the measured range but below the maximum of 10 kilonewtons, prompting the team to introduce a calibrated adjustment coefficient of 1.7 on the elastic modulus ratio to cover unfavorable stress states with a safety margin.

With the calibrated model in hand, the reinforcement’s contribution became quantifiable. Installing the mesh and truss shifts the neutral axis downward, shrinking the tension zone and cutting the peak tensile stress at the bottom edge, where cracks begin. The result is a 7.6 percent increase in initial cracking capacity, from 0.288 to 0.310 megapascals. More striking is the post-cracking behavior: at the moment the backfill first cracks, the steel carries only about 114 megapascals against a tensile strength of 375 megapascals, leaving substantial reserve. The model predicts the reinforced roof can still sustain 0.102 megapascals of overlying load after cracking, roughly 33 percent of its cracking capacity, with tensile rupture of the reinforcement, rather than crushing of the backfill, governing eventual failure.

Subidence calculations revealed a subtler picture. The classical elastic deflection formula for a simply supported beam predicted a mere 0.28 millimeters of sag, wildly short of the 16 millimeters measured in the field, exposing the inadequacy of ideal elastic theory for absolute prediction. A geometric alternative fared far better: by computing the elongation of the roof’s lower boundary under bending and approximating the sagging profile as a circular arc, the team obtained a theoretical subsidence of 31.5 millimeters, which a field-calibrated adjustment coefficient of 0.49 refined to 15.4 millimeters, closely matching observation. Before cracking, reinforcement barely changes subsidence at all. Its real value emerges afterward: by continuing to carry tensile force, it extends the allowable subsidence from 23.05 to 41.09 millimeters, an increase of about 78 percent, delaying crack propagation and instability rather than preventing initial deflection.

Finally, the team used the corrected model to optimize the reinforcement layout itself, holding truss geometry fixed and varying only bar diameters and spacings within a 200-to-500-millimeter window for the mesh. Evaluating total steel mass, cracking capacity, post-cracking capacity, and limiting subsidence across seven schemes, they identified a preferred configuration using 6-millimeter mesh bars at 200-millimeter spacing with trusses at 1000-millimeter spacing. Compared with the original design, this scheme raises the theoretical post-cracking load-bearing capacity by 22.5 percent while actually trimming material cost by 3.1 percent, with cracking capacity and deformation limits essentially unchanged. A cost-focused alternative cuts steel mass by 17 percent at unchanged performance. Because the schemes alter only spacing and diameter, they require no change to existing fabrication practice. The authors stress that both calibration coefficients are empirical values tied to Longshou Mine’s monitoring data and must be recalibrated elsewhere, and that hangers and other vertical components, whose load-transfer roles differ from horizontal tensile reinforcement, remain outside the model’s scope for future work. Even so, the study delivers something underground engineers have long lacked: a fast, transparent, field-tested formula connecting steel layout to roof survival.

Subject of Research: Mechanical modeling and field monitoring of steel-reinforced cemented backfill roofs in downward drift mining

Article Title: A mechanical model for bottom reinforcement strengthened backfill roofs in downward drifts revised by field monitoring

Article References: Ye, Y., Li, K., Hu, Y., Zhang, B., Guo, R., & Han, B. (2026). A mechanical model for bottom reinforcement strengthened backfill roofs in downward drifts revised by field monitoring. Results in Engineering, 32, Article 113264. https://doi.org/10.1016/j.rineng.2026.113264

Image Credits: AI Generated

DOI: 10.1016/j.rineng.2026.113264

Keywords: backfill roof, downward drift mining, cemented backfill, steel reinforcement, transformed section method, simply supported beam, field monitoring, roof subsidence, cracking capacity, post-cracking behavior, Longshou Mine, mining geotechnics